Tracking soil moisture dynamics with Vis-NIR spectroscopy during laboratory bare-soil evaporation
Numerous studies have investigated the relationship between spectral signal and soil moisture in the laboratory, usually using data from soil samples with predefined moisture levels for model calibration. However, it remains untested whether spectral monitoring can accurately capture the dynamic moi...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-07-01
|
| Series: | Geoderma |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S001670612500206X |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Numerous studies have investigated the relationship between spectral signal and soil moisture in the laboratory, usually using data from soil samples with predefined moisture levels for model calibration. However, it remains untested whether spectral monitoring can accurately capture the dynamic moisture changes occurring at the soil surface during drying. We conducted evaporation experiments on 5 cm tall packed soil columns of two soil types (sand and silt loam). The surface water content of each soil column was assessed by repeatedly recording spectra in the visible, near infrared, and shortwave infrared domain using an ASD Fieldspec® Pro spectrometer (350–2500 nm). The inferred water contents were then compared to those obtained from a numerical simulation with the Richards equation, which used soil hydraulic properties determined with the simplified evaporation method and measured evaporation rates as boundary condition. To develop the spectral model, samples with defined water contents were independently analyzed with the ASD. Among the three tested spectral models (polynomial linear regression, principal component regression (PCR) and partial least squares regression (PLSR)), the best model performance was achieved by polynomial linear regression. Regarding the transient evaporation experiments, the spectral model led to generally lower surface water contents than those predicted by the Richards equation. While soil moisture estimates for the silt loam closely matched simulated values (mean error = 2.81 vol%), the sandy soil exhibited systematic underestimations (mean error = 7.13 vol%), likely due to factors related to measurement setup and contact probe placement. |
|---|---|
| ISSN: | 1872-6259 |